[PATCH V4 0/3] handle polling erros in vhost/vhost_net

This is an update version of last version to fix the handling of polling errors
in vhost/vhost_net.

Currently, vhost and vhost_net ignore polling errors which can lead kernel
crashing when it tries to remove itself from waitqueue after the polling
failure. Fix this by:

- examing the POLLERR when set backend and report erros to userspace
- let tun always add to waitqueue in .poll() after the queue is created even if
  it was detached.

Changes from V3:
- make a smaller patch that doesn't touch the whole polling state and only check
  the polliner errors in backend setting.
- add a patch that allows tuntap to do polling/reading/writing when detached
  which could simplify the work of its user.

Changes from v2:
- check poll->wqh instead of the wrong assumption about POLLERR and waitqueue
- drop the whole tx polling state check since it was replaced by the wqh
  checking
- drop the buggy tuntap patch

Changes from v1:
- restore the state before the ioctl when vhost_init_used() fails
- log the error when meet polling errors in the data path
- don't put into waitqueue when tun_chr_poll() return POLLERR

Jason Wang (3):
  vhost_net: correct error handling in vhost_net_set_backend()
  vhost_net: handle polling errors when setting backend
  tuntap: allow polling/writing/reading when detached

 drivers/net/tun.c     |   18 +++++++++++-------
 drivers/vhost/net.c   |   41 ++++++++++++++++++++++++++++-------------
 drivers/vhost/vhost.c |   18 +++++++++++++++---
 drivers/vhost/vhost.h |    2 +-
 4 files changed, 55 insertions(+), 24 deletions(-)

Currently, the polling errors were ignored, which can lead following issues:

- vhost remove itself unconditionally from waitqueue when stopping the poll,
  this may crash the kernel since the previous attempt of starting may fail to
  add itself to the waitqueue
- userspace may think the backend were successfully set even when the polling
  failed.

Solve this by:

- check poll->wqh before trying to remove from waitqueue
- report polling errors in vhost_poll_start(), tx_poll_start(), the return value
  will be checked and returned when userspace want to set the backend

After this fix, there still could be a polling failure after backend is set, it
will addressed by the next patch.

Signed-off-by: Jason Wang <redacted>
---
 drivers/vhost/net.c   |   27 ++++++++++++++++++---------
 drivers/vhost/vhost.c |   18 +++++++++++++++---
 drivers/vhost/vhost.h |    2 +-
 3 files changed, 34 insertions(+), 13 deletions(-)

diff --git a/drivers/vhost/net.c b/drivers/vhost/net.c
index d10ad6f..959b1cd 100644
--- a/drivers/vhost/net.c
+++ b/drivers/vhost/net.c

@@ -165,12 +165,16 @@ static void tx_poll_stop(struct vhost_net *net)
 }
 
 /* Caller must have TX VQ lock */
-static void tx_poll_start(struct vhost_net *net, struct socket *sock)
+static int tx_poll_start(struct vhost_net *net, struct socket *sock)
 {
+	int ret;
+
 	if (unlikely(net->tx_poll_state != VHOST_NET_POLL_STOPPED))
-		return;
-	vhost_poll_start(net->poll + VHOST_NET_VQ_TX, sock->file);
-	net->tx_poll_state = VHOST_NET_POLL_STARTED;
+		return 0;
+	ret = vhost_poll_start(net->poll + VHOST_NET_VQ_TX, sock->file);
+	if (!ret)
+		net->tx_poll_state = VHOST_NET_POLL_STARTED;
+	return ret;
 }
 
 /* In case of DMA done not in order in lower device driver for some reason.

@@ -642,20 +646,23 @@ static void vhost_net_disable_vq(struct vhost_net *n,
 		vhost_poll_stop(n->poll + VHOST_NET_VQ_RX);
 }
 
-static void vhost_net_enable_vq(struct vhost_net *n,
+static int vhost_net_enable_vq(struct vhost_net *n,
 				struct vhost_virtqueue *vq)
 {
 	struct socket *sock;
+	int ret;
 
 	sock = rcu_dereference_protected(vq->private_data,
 					 lockdep_is_held(&vq->mutex));
 	if (!sock)
-		return;
+		return 0;
 	if (vq == n->vqs + VHOST_NET_VQ_TX) {
 		n->tx_poll_state = VHOST_NET_POLL_STOPPED;
-		tx_poll_start(n, sock);
+		ret = tx_poll_start(n, sock);
 	} else
-		vhost_poll_start(n->poll + VHOST_NET_VQ_RX, sock->file);
+		ret = vhost_poll_start(n->poll + VHOST_NET_VQ_RX, sock->file);
+
+	return ret;
 }
 
 static struct socket *vhost_net_stop_vq(struct vhost_net *n,

@@ -833,7 +840,9 @@ static long vhost_net_set_backend(struct vhost_net *n, unsigned index, int fd)
 		r = vhost_init_used(vq);
 		if (r)
 			goto err_used;
-		vhost_net_enable_vq(n, vq);
+		r = vhost_net_enable_vq(n, vq);
+		if (r)
+			goto err_used;
 
 		oldubufs = vq->ubufs;
 		vq->ubufs = ubufs;

diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c
index 34389f7..9759249 100644
--- a/drivers/vhost/vhost.c
+++ b/drivers/vhost/vhost.c

@@ -77,26 +77,38 @@ void vhost_poll_init(struct vhost_poll *poll, vhost_work_fn_t fn,
 	init_poll_funcptr(&poll->table, vhost_poll_func);
 	poll->mask = mask;
 	poll->dev = dev;
+	poll->wqh = NULL;
 
 	vhost_work_init(&poll->work, fn);
 }
 
 /* Start polling a file. We add ourselves to file's wait queue. The caller must
  * keep a reference to a file until after vhost_poll_stop is called. */
-void vhost_poll_start(struct vhost_poll *poll, struct file *file)
+int vhost_poll_start(struct vhost_poll *poll, struct file *file)
 {
 	unsigned long mask;
+	int ret = 0;
 
 	mask = file->f_op->poll(file, &poll->table);
 	if (mask)
 		vhost_poll_wakeup(&poll->wait, 0, 0, (void *)mask);
+	if (mask & POLLERR) {
+		if (poll->wqh)
+			remove_wait_queue(poll->wqh, &poll->wait);
+		ret = -EINVAL;
+	}
+
+	return ret;
 }
 
 /* Stop polling a file. After this function returns, it becomes safe to drop the
  * file reference. You must also flush afterwards. */
 void vhost_poll_stop(struct vhost_poll *poll)
 {
-	remove_wait_queue(poll->wqh, &poll->wait);
+	if (poll->wqh) {
+		remove_wait_queue(poll->wqh, &poll->wait);
+		poll->wqh = NULL;
+	}
 }
 
 static bool vhost_work_seq_done(struct vhost_dev *dev, struct vhost_work *work,

@@ -792,7 +804,7 @@ long vhost_vring_ioctl(struct vhost_dev *d, int ioctl, void __user *argp)
 		fput(filep);
 
 	if (pollstart && vq->handle_kick)
-		vhost_poll_start(&vq->poll, vq->kick);
+		r = vhost_poll_start(&vq->poll, vq->kick);
 
 	mutex_unlock(&vq->mutex);
 

diff --git a/drivers/vhost/vhost.h b/drivers/vhost/vhost.h
index 2639c58..17261e2 100644
--- a/drivers/vhost/vhost.h
+++ b/drivers/vhost/vhost.h

@@ -42,7 +42,7 @@ void vhost_work_queue(struct vhost_dev *dev, struct vhost_work *work);
 
 void vhost_poll_init(struct vhost_poll *poll, vhost_work_fn_t fn,
 		     unsigned long mask, struct vhost_dev *dev);
-void vhost_poll_start(struct vhost_poll *poll, struct file *file);
+int vhost_poll_start(struct vhost_poll *poll, struct file *file);
 void vhost_poll_stop(struct vhost_poll *poll);
 void vhost_poll_flush(struct vhost_poll *poll);
 void vhost_poll_queue(struct vhost_poll *poll);

-- 
1.7.1

We forbid polling, writing and reading when the file were detached, this may
complex the user in several cases:

- when guest pass some buffers to vhost/qemu and then disable some queues,
  vhost/qemu needs to to its own cleanup which is complex. We can do this simply
  by allowing a user can still write to an disabled queue to handle this. And
  user can still do read but just nothing returned.
- align the polling behavior with macvtap which never fails when the queue is
  created. this can simplify the polling errors handling of its user (e.g vhost)

Signed-off-by: Jason Wang <redacted>
---
 drivers/net/tun.c |   18 +++++++++++-------
 1 files changed, 11 insertions(+), 7 deletions(-)

diff --git a/drivers/net/tun.c b/drivers/net/tun.c
index af372d0..eb68937 100644
--- a/drivers/net/tun.c
+++ b/drivers/net/tun.c

@@ -307,7 +307,7 @@ static void tun_flow_update(struct tun_struct *tun, u32 rxhash,
 
 	rcu_read_lock();
 
-	if (tun->numqueues == 1)
+	if (tun->numqueues == 1 || queue_index >= tun->numqueues)
 		goto unlock;
 
 	e = tun_flow_find(head, rxhash);

@@ -406,21 +406,21 @@ static void __tun_detach(struct tun_file *tfile, bool clean)
 
 	tun = rtnl_dereference(tfile->tun);
 
-	if (tun) {
+	if (tun && !tfile->detached) {
 		u16 index = tfile->queue_index;
 		BUG_ON(index >= tun->numqueues);
 		dev = tun->dev;
 
 		rcu_assign_pointer(tun->tfiles[index],
 				   tun->tfiles[tun->numqueues - 1]);
-		rcu_assign_pointer(tfile->tun, NULL);
 		ntfile = rtnl_dereference(tun->tfiles[index]);
 		ntfile->queue_index = index;
 
 		--tun->numqueues;
-		if (clean)
+		if (clean) {
+			rcu_assign_pointer(tfile->tun, NULL);
 			sock_put(&tfile->sk);
-		else
+		} else
 			tun_disable_queue(tun, tfile);
 
 		synchronize_net();

@@ -465,6 +465,10 @@ static void tun_detach_all(struct net_device *dev)
 		rcu_assign_pointer(tfile->tun, NULL);
 		--tun->numqueues;
 	}
+	list_for_each_entry(tfile, &tun->disabled, next) {
+		wake_up_all(&tfile->wq.wait);
+		rcu_assign_pointer(tfile->tun, NULL);
+	}
 	BUG_ON(tun->numqueues != 0);
 
 	synchronize_net();

@@ -491,7 +495,7 @@ static int tun_attach(struct tun_struct *tun, struct file *file)
 	int err;
 
 	err = -EINVAL;
-	if (rtnl_dereference(tfile->tun))
+	if (rtnl_dereference(tfile->tun) && !tfile->detached)
 		goto out;
 
 	err = -EBUSY;

@@ -1795,7 +1799,7 @@ static int tun_set_queue(struct file *file, struct ifreq *ifr)
 			ret = tun_attach(tun, file);
 	} else if (ifr->ifr_flags & IFF_DETACH_QUEUE) {
 		tun = rtnl_dereference(tfile->tun);
-		if (!tun || !(tun->flags & TUN_TAP_MQ))
+		if (!tun || !(tun->flags & TUN_TAP_MQ) || tfile->detached)
 			ret = -EINVAL;
 		else
 			__tun_detach(tfile, false);

-- 
1.7.1

Currently, when vhost_init_used() fails the sock refcnt and ubufs were
leaked. Correct this by calling vhost_init_used() before assign ubufs and
restore the oldsock when it fails.

Signed-off-by: Jason Wang <redacted>
---
 drivers/vhost/net.c |   16 +++++++++++-----
 1 files changed, 11 insertions(+), 5 deletions(-)

diff --git a/drivers/vhost/net.c b/drivers/vhost/net.c
index ebd08b2..d10ad6f 100644
--- a/drivers/vhost/net.c
+++ b/drivers/vhost/net.c

@@ -827,15 +827,16 @@ static long vhost_net_set_backend(struct vhost_net *n, unsigned index, int fd)
 			r = PTR_ERR(ubufs);
 			goto err_ubufs;
 		}
-		oldubufs = vq->ubufs;
-		vq->ubufs = ubufs;
+
 		vhost_net_disable_vq(n, vq);
 		rcu_assign_pointer(vq->private_data, sock);
-		vhost_net_enable_vq(n, vq);
-
 		r = vhost_init_used(vq);
 		if (r)
-			goto err_vq;
+			goto err_used;
+		vhost_net_enable_vq(n, vq);
+
+		oldubufs = vq->ubufs;
+		vq->ubufs = ubufs;
 
 		n->tx_packets = 0;
 		n->tx_zcopy_err = 0;

@@ -859,6 +860,11 @@ static long vhost_net_set_backend(struct vhost_net *n, unsigned index, int fd)
 	mutex_unlock(&n->dev.mutex);
 	return 0;
 
+err_used:
+	rcu_assign_pointer(vq->private_data, oldsock);
+	vhost_net_enable_vq(n, vq);
+	if (ubufs)
+		vhost_ubuf_put_and_wait(ubufs);
 err_ubufs:
 	fput(sock->file);
 err_vq:

-- 
1.7.1

On Tue, Jan 15, 2013 at 06:28:46PM +0800, Jason Wang wrote:

Hmm I don't understand - does something ensure that queue_index is >
numqueues if the queue is disabled?
Can we check tun->disabled and skip tun_flow_update completely?
We might need to switch to rcu_assign for tun->disabled for this
but otherwise it looks easy.

On Tuesday, January 15, 2013 06:48:08 PM Michael S. Tsirkin wrote:

On Tue, Jan 15, 2013 at 06:28:46PM +0800, Jason Wang wrote:

quoted

We forbid polling, writing and reading when the file were detached, this
may complex the user in several cases:

- when guest pass some buffers to vhost/qemu and then disable some queues,

  vhost/qemu needs to to its own cleanup which is complex. We can do this
  simply by allowing a user can still write to an disabled queue to
  handle this. And user can still do read but just nothing returned.

- align the polling behavior with macvtap which never fails when the queue
is> 
  created. this can simplify the polling errors handling of its user (e.g
  vhost)> 
Signed-off-by: Jason Wang <redacted>
---

 drivers/net/tun.c |   18 +++++++++++-------
 1 files changed, 11 insertions(+), 7 deletions(-)

diff --git a/drivers/net/tun.c b/drivers/net/tun.c
index af372d0..eb68937 100644
--- a/drivers/net/tun.c
+++ b/drivers/net/tun.c

@@ -307,7 +307,7 @@ static void tun_flow_update(struct tun_struct *tun,

u32 rxhash,> 
 	rcu_read_lock();

-	if (tun->numqueues == 1)
+	if (tun->numqueues == 1 || queue_index >= tun->numqueues)

 		goto unlock;
 	
 	e = tun_flow_find(head, rxhash);

Hmm I don't understand - does something ensure that queue_index is >
numqueues if the queue is disabled?

I miss this, something is needed.

Can we check tun->disabled and skip tun_flow_update completely?
We might need to switch to rcu_assign for tun->disabled for this
but otherwise it looks easy.

Correct, will post another version, thanks.

quoted

@@ -406,21 +406,21 @@ static void __tun_detach(struct tun_file *tfile,

bool clean)> 
 	tun = rtnl_dereference(tfile->tun);

-	if (tun) {
+	if (tun && !tfile->detached) {

 		u16 index = tfile->queue_index;
 		BUG_ON(index >= tun->numqueues);
 		dev = tun->dev;
 		
 		rcu_assign_pointer(tun->tfiles[index],
 		
 				   tun->tfiles[tun->numqueues - 1]);

-		rcu_assign_pointer(tfile->tun, NULL);

 		ntfile = rtnl_dereference(tun->tfiles[index]);
 		ntfile->queue_index = index;
 		
 		--tun->numqueues;

-		if (clean)
+		if (clean) {
+			rcu_assign_pointer(tfile->tun, NULL);

 			sock_put(&tfile->sk);

-		else
+		} else

 			tun_disable_queue(tun, tfile);
 		
 		synchronize_net();

@@ -465,6 +465,10 @@ static void tun_detach_all(struct net_device *dev)

 		rcu_assign_pointer(tfile->tun, NULL);
 		--tun->numqueues;
 	
 	}

+	list_for_each_entry(tfile, &tun->disabled, next) {
+		wake_up_all(&tfile->wq.wait);
+		rcu_assign_pointer(tfile->tun, NULL);
+	}

 	BUG_ON(tun->numqueues != 0);
 	
 	synchronize_net();

@@ -491,7 +495,7 @@ static int tun_attach(struct tun_struct *tun, struct

file *file)> 
 	int err;
 	
 	err = -EINVAL;

-	if (rtnl_dereference(tfile->tun))
+	if (rtnl_dereference(tfile->tun) && !tfile->detached)

 		goto out;
 	
 	err = -EBUSY;

@@ -1795,7 +1799,7 @@ static int tun_set_queue(struct file *file, struct

ifreq *ifr)> 
 			ret = tun_attach(tun, file);
 	
 	} else if (ifr->ifr_flags & IFF_DETACH_QUEUE) {
 	
 		tun = rtnl_dereference(tfile->tun);

-		if (!tun || !(tun->flags & TUN_TAP_MQ))
+		if (!tun || !(tun->flags & TUN_TAP_MQ) || tfile->detached)

 			ret = -EINVAL;
 		
 		else
 		
 			__tun_detach(tfile, false);

--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html